Method for engineering strand-specific, sequence-specific, DNA-nicking enzymes

Abstract

Methods are provided for converting into a sequence specific strand specific and location specific DNA nicking endonuclease, a restriction endonuclease that recognizes an asymmetric DNA sequence, the endonuclease having two catalytic sites and one or more single sequence specific DNA-binding domains. In one embodiment the method requires inactivating one of the catalytic sites of the restriction endonuclease. In another embodiment, the restriction endonuclease is a dimer having a first and second subunit each comprising a sequence specific DNA binding domain, a catalytic site and a dimerization domain. The nicking endonuclease is formed from combining one subunit having an inactivated catalytic site and a second subunit having an inactivated DNA binding domain. The nicking endonuclease may be converted into a restriction endonuclease by the addition of manganese cations in the digestion buffer.

Description

RELATED APPLICATIONS

[0001] This is a Non-Provisional Application of Provisional Application Serial No. 60/314,386 filed on Aug. 23, 2001 herein incorporated by reference.

BACKGROUND OF THE INVENTION

[0002] Enzymes are fundamental components of all living cells. Encoded by genes, and comprising proteins for the most part, enzymes orchestrate the multitude of chemical processes upon which life depends. Simple living organisms, such as bacteria, synthesize around 1000 enzymes, the minimum repertoire for independent existence, while complex organisms such as ourselves synthesize far more, perhaps as many as 30,000. Enzymes are profoundly powerful; the biochemical nuances of a single enzyme among thousands can spell the difference between health or disease; survival or death.

[0003] There is considerable interest in enzymes, both as objects for study, and as molecular tools for experimentation and analysis in the laboratory. Chief among the latter are restriction endonucleases, enzymes synthesized by bacteria and archae that catalyze the cleavage of double-stranded DNA molecules. Over two hundred and fifty distinct restriction endonucleases have been discovered, each differing in the specifics of the sites within DNA at which they induce cleavage (Roberts and Macelis, Nucleic Acids Res. 28:306-307 (2000)). Two hundred of these are manufactured commercially, and many have become tools of the trade for molecular biologists in scientific and medical institutions around the world.

[0004] Restriction endonucleases bind to duplex DNA at particular sequences of nucleotides termed ‘recognition sequences’. Once bound, they hydrolyze the two sugar-phosphate strands that make up the paired, helical backbones of the DNA molecule. Following hydrolysis, the endonucleases generally detach from the DNA, and the two fragments of the severed DNA molecule are released. As a rule, the two strand-hydrolysis reactions proceed in parallel in a double reaction that requires the presence of two catalytic sites within each enzyme, one for hydrolyzing each strand. In most cases, these two catalytic sites are identical because, in their catalytically active forms at the moment of DNA cleavage, most restriction endonucleases are homodimeric; that is to say, they are composed of two identical protein subunits, or assemblages of subunits, bound to each other in symmetrical, if fleeting, opposition. Each subunit or assemblage possesses a single catalytic site, and so the active homodimer possesses two.

[0005] For certain laboratory applications it would be useful if variant restriction endonucleases were available that hydrolyzed only one strand of duplex DNA rather than hydrolyzing both strands. Because such enzymes—‘DNA-nicking’ endonucleases as opposed to ‘DNA-cleaving’ endonucleases—are uncommon in nature, we have developed a method for producing them in the laboratory. This method, and the nicking endonucleases we have made, or anticipate making, by application of the method form the basis for this patent application. The method relies on identifying a restriction endonuclease that possesses two different catalytic sites for strand hydrolysis, rather than two identical sites, and then inactivating one site or the other by mutation to form an altered enzyme in which only one of the two sites retains catalytic ability. Provided the inactivating mutation(s) do not interfere with the functioning of the other catalytic site, or with the ability of the protein to fold properly and bind to its recognition sequence in DNA, the mutated enzyme should hydrolyze only one strand of DNA. Nicking enzymes created in this way will be sequence-specific, in that nicking will occur only at the recognition sequence. They will be strand-specific, in that which of the two DNA strands becomes nicked will depend upon which of the two catalytic sites in the enzyme retains function. And they will be position-specific in that the position of the nick with respect to the recognition sequence, determined by the geometry of the enzyme, will be predictable and constant.

SUMMARY OF THE INVENTION

[0006] A method is provided whereby certain restriction endonucleases that hydrolyze both strands of duplex DNA (‘cleave’) can be converted into enzymes that hydrolyze only one strand of duplex DNA (‘nick’), instead. The method involves identifying a restriction endonuclease that possesses two different catalytic sites for strand-hydrolysis, and then discretely inactivating one site or the other by mutation to create a variant enzyme that possesses only one catalytic site. Nicking enzymes created in this way hydrolyze only one strand of DNA, and they do so in a sequence-specific, strand-specific, and location-specific manner. The nicks introduced into DNA by the actions of these enzymes can serve as initiation points for further enzymatic reactions including strand-specific DNA amplification, degradation, mutagenesis, polymerization, and recombination. The restriction endonucleases to which this method is applicable are primarily those that recognize continuous asymmetric nucleotide sequences, and cleave within or close to one side of these sequences; enzymes, that is to say, belonging to the Type IIs, IIt, and IIf classes. (New England Biolabs nomenclature, 2000) We applied the method to R.BbvCI, a Type IIt restriction endonuclease that cleaves DNA within the asymmetric seven-base pair recognition sequence 5′-CCTCAGC-3′. The method generated a novel pair of complimentary nicking enzymes, Nt.BbvCI, that nicked DNA two bases inside the ‘top’ strand of the sequence, thus: 5′-CCˆ TCAGC-3′; and Nb.BbvCI, that nicked DNA two bases inside the ‘bottom’ strand, thus: 5′-GCˆ TGAGG-3′ (see Example 1).

[0007] The method described here is applicable to the Type IIs endonucleases including R.BbvI and N.BstNBI, and provides a means to generate additional novel endonucleases that nick either DNA strand several bases downstream of the sequences 5′-GCAGC-3′ and 5′-GAGTC-3′, respectively (Example 2). A variation of the method is applicable to other Type II endonucleases such as R.FokI (Type IIs), and to certain Type IIf endonuclease-methyltransferases, such as Eco57I, that do not normally use two different catalytic sites to accomplish DNA cleavage, but might be made to do so by genetic manipulation (Example 3).

BRIEF DESCRIPTION OF THE DRAWINGS

[0008] FIG. 1. Nucleotide sequence of the BbvCI restriction and modification genes (SEQ ID NO: 1).

[0009] FIG. 2. Amino acid sequences of the BbvCI methyltransferases, and endonuclease subunits (SEQ ID NO: 2, SEQ ID NO: 3, SEQ ID NO: 4 and SEQ ID NO: 5).

[0010] FIG. 3. Nucleotide sequence of the BlpI restriction and modification genes (SEQ ID NO: 6).

[0011] FIG. 4. Amino acid sequence of the BlpI methyltransferase and endonuclease (SEQ ID NO: 7 and SEQ ID NO: 8).

[0012] FIG. 5. Nucleotide sequence of the Bsu36I restriction and modification genes (SEQ ID NO: 9).

[0013] FIG. 6. Amino acid sequence of the BSu36I methyltransferase and endonuclease (SEQ ID NO: 10 and SEQ ID NO: 11).

[0014] FIG. 7. Amino acid sequence alignment between the BlpI (SEQ ID NO: 8), Bsu36I (SEQ ID NO: 11), and DdeI (SEQ ID NO: 12) restriction endonucleases and the two endonuclease subunits of BbvCI (SEQ ID NO: 4 and SEQ ID NO: 5). DdeI differs so radically from the other four proteins beyond the catalytic site regions (boxes) that alignment has only been attempted with DdeI in these regions. Regions of marked similarity among the other four proteins are shown in red. Particularly well-conserved residues in these regions are shown in bold.

[0015] FIG. 8. Endonucleolytic activity assays of mutant enzymes. Crude extracts of the enzymes were incubated with supercoiled pUB DNA, and then electrophoresed through 1% agarose. Lane 1 (far left) contains molecular-weight markers. Lanes 2-9 contain the digestion products of enzymes 1-33 to 1-40. Control lanes 10 and 11 contain wild type R.BbvCI-digested plasmid, and undigested plasmid, respectively. Lanes 12-19 contain the digestion products of enzymes 2-33 to 2-40. Certain enzymes (e.g. 1-38, 2-39; filled black circles) cleave the plasmid. Other enzymes (e.g.1-35, 2-37; filled black squares) nick it. And yet others (e.g. 1-36, 2-34) appear to be inactive.

[0016] FIG. 9. SDS-PAGE analysis of purified wild type R.BbvCI restriction endonuclease (lanes 2-3: prep. ‘Dan’; lanes 4-6: prep. ‘Diane’), and of purified N.BbvCI nicking endonucleases 2-12 (lanes 7-9), and 1-37 (lanes 10-12). Lane 1 (far left) contains molecular weight markers. All the enzymes preps migrate as two bands of roughly equal intensity signifying that the enzymes are heterodimers with 1:1 subunit stoichiometry.

[0017] FIG. 10. High-resolution agarose gel electrophoresis of supercoiled pUB DNA titrated with varying amounts of restriction endonucleases R.BbvCI (lanes 2-6) and R.EcoRI (lanes 7-11), and nicking endonuclease N.BbvCI 1-37 (lanes 11-15). Lane 1 (far left) contains molecular weight markers. Lane 16 contains undigested pUB substrate DNA. EcoRI (1 site) cleaves pUB to produce a full-length linear fragment. R.BbvCI (2 sites) produces full-length linear fragment transiently, but this is cleaved further into two fragments. (Only the larger of these two fragments is visible; the smaller fragment migrated beyond the bottom of the gel.) N.BbvCI 1-37 nicks pUB to produce a full length open-circular product that migrates slightly slower than its linear counterpart.

[0018] FIG. 11. Agarose gel of supercoiled pUB DNA nicked in sequence with nicking enzymes N.BbvCI 1-37 and 2-12. Lane 1 (far left; R.BbvCI-digested pUB), and lanes 10 (N.BbvCI-nicked pUB) and 11 (undigested pUB) are controls. pUB DNA was nicked first with N.BbvCI 1-37 and then purified by microfiltration (lane 9). This nicked DNA was then redigested with more N.BbvCI 1-37 (lane 8), or with N.BbvCI 2-12 (lane 7) or R.BbvCI (lane 6). Alternately, pUB DNA was first nicked with N.BbvCI 2-12 and purified (lane 5), and then redigested with N.BbvCI 1-37 (lane 4), more N.BbvCI 2-12 (lane 3), or with R.BbvCI. Digesting with either nicking enzyme alone nicks the DNA. Redigesting with the same enzyme has no further affect, but redigesting with the other enzyme results in double-strand cleavage identical to that produced by R.BbvCI.

[0019] FIG. 12. Automated DNA sequencing traces in both directions through one of the BbvCI sites in pUB DNA digested with nicking enzymes N.BbvCI 1-37 or 2-12. Nicking with N.BbvCI 1-37 (top traces) does not affect the ‘forward’ sequencing reaction, but it causes the ‘reverse’ reaction to terminate at the BbvCI site. This demonstrates that enzyme 1-37 (composition: R1−:R2+) nicks the ‘top’, CCTCAGC, strand of the recognition sequence and not the ‘bottom’ strand. Nicking with N.BbvCI 2-12 has the opposite effect (bottom traces), demonstrating that this enzyme (composition: R1+:R2−) nicks the bottom, GCTCAGG, strand of the recognition sequence and not the top strand.

[0020] FIG. 13. Application of method for conversion to nicking endonucleases of the restriction endonuclease R.BbvCI. R.BbvCI is composed of two different subunits, R1 and R2, each of which contains one catalytic site for DNA strand-hydrolysis. R.BbvCI (composition: R1+:R2+) cleaves duplex DNA at its recognition site by hydrolyzing the two strands of the duplex simultaneously, one catalytic site inducing cleavage of each strand. In turn, each catalytic site in R.BbvCI was inactivated to produce two novel nicking endonuclease of composition R1−:R2+ and R1+:R2−.

[0021] FIG. 14. SDS-PAGE analysis of whole-cell proteins of isolates synthesizing mutant BbvCI enzymes. The two subunits of the enzyme can be easily distinguished. Cells that exhibit R.BbvCI endonuclease activity in crude extracts synthesize both subunits (e.g. Wt; 1-5). Mutant cells that exhibit no endonuclease activity in crude extracts either lack one of the subunits (e.g. 1-2; 2-8), or synthesize subunits that contain incapacitating mutations.

[0022] FIG. 15. Restriction endonucleases likely to possess two different catalytic sites for DNA strand-hydrolysis. These enzymes, and others like them, are candidates for conversion to DNA-nicking enzymes by the present invention. These enzymes recognize continuous asymmetric DNA sequences and catalyze strand-hydrolysis within those sequences or just outside. Many of the enzymes come from bacteria of the genus Bacillus, suggesting that this genus could be a productive source of further enzymes of this category.

[0023] FIG. 16. Endonuclease yields and activities from 100 gm wet pastes of cells expressing wild type BbvCI, and the nicking derivatives 1-37 and 2-12. 1 unit of R.BbvCI activity is defined here as the least amount of enzyme that completely converts 1 microgram of supercoiled pUB DNA (2 recognition sites) into two fragments in 1 h at 37° C. in 50 microL of NEBuffer 4 (50 mM potassium acetate, 20 mM Tris acetate, 10 mM magnesium acetate, 1 mM dithiothreitol, pH 7.9) plus 100 microgram/ml BSA. 1 unit of N.BbvCI activity is the least amount of enzyme that completely converts 1 microgram of supercoiled pUB DNA into the open-circular form in 1 h at 37° C. in 50 microL in the same buffer. These enzymes were homogeneous by SDS-PAGE, and had a final protein concentration of approximately 40 mg/ml.

[0024] FIG. 17. Properties of enzymes with mutated R1-subunit catalytic sites. The mutant isolate number is given in column 1. The nt. sequence of the section of the R1 gene coding for the subunit-1 catalytic site in each isolate is given in column 2 (SEQ ID NO: 13 through SEQ ID NO: 35), and the amino acid sequence is given in column 3. In all of the isolates, subunit 1 was mutated at the active site, and subunit 2 was normal (‘+’ in column 4). The catalytic activities of crude cell extracts of these enzymes is shown in column 5-7. A ‘✓’ in column 5 signifies that the enzyme cleaves DNA (e.g. 1-5); a ‘✓’ in column 6 signifies that it nicks DNA (e.g. 1-4); and a ‘✓’ in column 7 signifies that it is inactive and neither cleaves DNA nor nicks it (e.g. 1-14). A ‘(✓)’ in column 6 indicates that the enzyme displays very weak nicking activity (e.g. 1-45). The symbol ‘mn’ in column 5 signifies that the enzyme cleaves DNA in the presence of 10 mM manganese chloride (e.g. 1-29).

[0025] FIG. 18. Properties of enzymes with mutated R2-subunit catalytic sites. The mutant isolate no. is given in column 1. The nt. sequence of the section of the R2 gene coding for the subunit-2 catalytic site in each isolate is given in column 2 (SEQ ID NO: 36 through SEQ ID NO: 63), and the amino acid sequence is given in column 4. In all of the isolates, subunit 2 was mutated at the active site, and subunit 1 was normal (‘+’ in column 3). The catalytic activities of crude cell extracts of these enzymes is shown in column 5-7. A ‘✓’ in column 5 signifies that the enzyme cleaves DNA (e.g. 2-39); a ‘✓’ in column 6 signifies that it nicks DNA (e.g. 2-2); and a ‘✓’ in column 7 signifies that it is inactive and neither cleaves DNA nor nicks it (e.g. 2-5). A ‘(✓)’ in column 6 indicates that the enzyme displays very weak nicking activity (e.g. 2-49). The symbol ‘mn’ in column 5 signifies that the enzyme cleaves DNA in the presence of (e.g. 2-56).

DETAILED DESCRIPTION OF INVENTION

[0026] In its preferred embodiment, the method involves the following steps. 1) Identifying a restriction endonuclease that possesses two different catalytic sites. 2) Cloning and sequencing the gene(s) for that endonuclease, together, if necessary, with the gene(s) for protective modification enzyme(s). 3) Identifying the amino acid residues within the endonuclease that form at least one of the catalytic sites for DNA strand-hydrolysis. 4) Altering one or more of the catalytic site residues by changing the nucleotides that code for them in the endonuclease gene(s). 5) Isolating clones that express altered restriction endonucleases, and identifying those that remain active, but now nick DNA at the recognition sequence rather than cleave it. These steps are described in detail below.

[0027] 1. Identifying Endonucleases with Two Different Catalytic Sites

[0028] Over three thousand restriction endonucleases have been discovered and characterized from a wide variety of bacteria and archae. Comprehensive lists of their recognition sequences and cleavage sites can be found at REBASE (http://rebase.neb.com). In the majority of restriction enzymes, the two catalytic sites are likely to be identical, because the enzymes either bind to DNA as homodimers, or they bind with some other organization but form homodimers transiently to accomplish DNA cleavage. Certain enzymes however possess two different catalytic sites and do not homodimerize.

[0029] The kinds of nucleotide sequences to which proteins bind reflects their organizations. Proteins that bind to symmetric (‘palindromic’) sequences do so because they, themselves, are symmetric (e.g. EcoRI: GAATTC). Proteins that bind to asymmetric nucleotide sequences, conversely, do so because they are asymmetric (e.g. FokI: GGATG). Since restriction enzymes that possess two different catalytic sites are generally asymmetric, it follows that the sequences to which they will bind will be asymmetric to match. The first step in identifying endonucleases likely to possess two different catalytic sites, then, is to eliminate from the list of all enzymes those that recognize symmetric nucleotide sequences. The next step is to eliminate enzymes that probably homodimerize to accomplish DNA cleavage. This includes the Type I endonuclease-methyltransferases (e.g. EcoKI: AACNNNNNNGTGC), and the Type IIf endonuclease-methyltransferases (also referred to as Type IV; e.g. BcgI: CGANNNNNNTGC) that recognize discontinuous asymmetric sequences. Eliminated, too, are the Type III endonucleases (e.g. EcoP1I: AGACC) that recognize continuous asymmetric sequences but cleave only when two such, opposed, sequences are present in the DNA (Humbelin et al., 3. Mol. Biol. 200:23-29 (1988); Meisel et al., Nature. 355:467-469 (1992)). However Type IIs endonucleases for which experimental evidence of homodimerization exists may be modified to provide nicking endonucleases (Example 3).

[0030] When endonucleases bind to DNA asymmetrically but cleave by homodimerization, only one component of the homodimer binds to the local recognition sequence. The other component contributes the second catalytic site and some non-specific binding energy, but it is thought to perform no nucleotide sequence-discrimination. The greater the separation between the recognition sequence and the hydrolysis sites on the DNA molecule, the easier it is to envisage this happening. When the hydrolysis sites are within the recognition sequence, or just outside, it seems unlikely that the enzyme would be symmetric, overall, but rather that it would be asymmetric, and would thus possess two different catalytic sites. We consider restriction endonucleases that recognize continuous, asymmetric sequences, and cleave within those sequences or very close to them, to be the ones most likely to possess two different catalytic sites. These enzymes are variously referred to as Type IIt or Type IIq endonucleases, or as atypical Type IIs endonucleases (Kessler and Holtke, Gene. 47:1-153 (1986); Degtyarev et al., Nucleic Acids Res. 18:5807-5810 (1990); Stankevicius et al., Nucleic Acids Res. 26:1084-1091 (1998); Degtyarev et al., Nucleic Acids Res. 28:e56 (2000)).

[0031] Thirteen kinds of restriction endonucleases currently fall into this category (FIG. 15). Seven—represented by AciI, BbvCI, Bpu10I, BsrBI, BssSI, BtrI, and GdiII—hydrolyze both strands of DNA within their recognition sequences, and these seem to us to be the most promising candidates of all for conversion to nicking enzymes by the present invention. Three more—represented by BsmI, BsrI, and SimI—hydrolyze one strand within the recognition sequence and one strand just outside, and the remaining three—represented by BsrDI, BstF5I, and BtsI—hydrolyze both strands just outside the recognition sequence. These last six kinds of endonucleases are also promising candidates for conversion to nicking enzymes. Among endonucleases for which the separation between the recognition sequence and hydrolysis sites is substantial are two, BbvI and BstNBI, that possibly act as heterodimers (see Example 2). If they indeed act this way, then they and other Type IIs endonucleases like them are also candidates for conversion to nicking enzymes.

[0032] Enzymes are composed of functional parts: one part binds a substrate, for example, another binds a co-factor, and a third catalyzes a chemical transformation. These parts associate with one another in the final enzyme but whether they do so as separate subunits or as interconnected domains of a larger, composite, protein does not, in principal, affect the outcome. In the course of evolution, genes encoding proteins that interact often migrate to adjacent loci in the genome. Sometimes, these neighboring genes then fuse, head-to-tail, to form one continuous gene that incorporates the ancestral proteins into a single polypeptide chain. When this happens, what was originally a multimeric protein of different subunits becomes a monomeric protein of several domains.

[0033] Restriction endonucleases that are catalytically asymmetric are likely to act either as heterodimers, comprising two different protein chains, or monomers, comprising a single protein chain. Bpu10I (FIG. 15) is reported to be a heterodimer (Stankevicius et al., Biologija. 2:51-53 (1996)). BbvCI, we show here, is also a heterodimer (see Example 1). BstNBI and BbvI might be heterodimers, too. In contrast, AciI, BsrI, BsrBI, and BssSI, are monomers. In the compact, Type IIt, variety of heterodimer, both subunits will likely perform nucleotide sequence recognition as well as catalysis, and so the subunits of this kind of enzyme will likely be of similar size, approximately 200-350 amino acids (aa) apiece. In contrast, in the extended, Type IIs, variety of heterodimer, both subunits will likely perform catalysis, but only one will perform sequence recognition, and consequently the subunits will differ in size, the small subunit being approximately 150-250 aa, the other being 2-3 times larger, or approximately 400-650 aa. Although the true subunit stoichiometry in heterodimers is 1:1, subunit dissociation can occur during enzyme purification resulting in a different apparent subunit stoichiometry in purified enzyme preparations.

[0034] The genes for the two subunits of heterodimeric restriction endonucleases will likely lie next to each other in the genomes, and in the same orientation. For Type IIt enzymes, the two genes will be similar in length. For Type IIs enzymes they will be different, one being 2-3 times longer than the other. Gene fusion could join the genes for the two subunits together, and so these endonucleases could also exist as monomers. The molecular masses of such monomers would be roughly the sum of the masses of the ancestral subunits, that is to say between approximately 450 and 900 aa. The specific activity of such monomers would be in the range typical for Type II and Type IIs enzymes, and considerably higher than those of partially dissociated heterodimers.

[0035] Heterodimeric restriction endonucleases may be recognized (i) by analysis of endonuclease purified from the original organism or from a recombinant host containing the cloned restriction system, or (ii) by sequence analysis of the cloned restriction system. In the former case, the purified endonuclease may be characterized by electrophoresis on SDS-PAGE, which will usually reveal the presence of two protein components migrating at different positions. It may be the case that the two subunits, although distinct in sequence and the products of different genes, will migrate at the same mobility on SDS-PAGE. This situation can be recognized because the apparent molecular weight derived from SDS-PAGE analysis will be one-half of the apparent molecular weight derived from gel-filtration analysis. Further, N-terminal amino acid sequence analysis of the purified endonuclease will reveal the presence of two different amino acids at each sequencing cycle, in the apparently single band. Procedures for determining these properties are well known in the art, and are disclosed for example in Current Protocols in Protein Analysis (sections 8.3, 10.1, and 11.10; Coligan, F. E., Dunn, B. M., Ploegh, H. L., Speicher, D. W., and Wingfield, P. T. Current Protocols in Protein Science, John Wiley and Sons, (1997)).

[0036] In the latter case, the restriction systems amenable to this invention will contain up to four open reading frames (ORFs), two encoding methyltransferases, and two encoding the subunits of the restriction endonuclease. The methyltransferase ORFs may be recognized by amino acid sequence comparisons according to (Wilson, Methods Enzymol. 216:259-279 (1992) and Malone, et al., J. Mol. Biol. 253:618-632 (1995)). Additional ORFs may also be present including those encoding proteins involved in the regulation of gene expression (such as C proteins), and in the repair of base-mismatches resulting from the deamination of 5-methylcytosine (such as Vsr proteins).

[0037] Genes encoding subunits of the endonuclease may be verified by creating expression clones in which the methyltransferase genes are carried on one plasmid, and the candidate endonuclease genes are carried on one or more additional plasmid(s), as disclosed in Brooks, et al. (U.S. Pat. No. 5,320,957). The presence of just the methyltransferase plasmid will cause the DNA within the cell to become resistant to action of the endonuclease, but the cells will express no endonuclease activity. Addition of the endonuclease genes on the additional plasmid(s) will cause the cells to express endonuclease activity. In some situations it may be possible to express the endonuclease genes in the absence of the methyltransferase genes, as disclosed in International Publication No. WO 99/11821.

[0038] The requirement for both ORFs for endonuclease activity of heterodimeric enzymes may be verified by creating expression clones in which each of the ORFs can be expressed separately, e.g. by placing each ORF on a separate compatible plasmid, or by placing each ORF under the control of a promoter that can be induced separately (e.g. inducible by lactose or by arabinose), and then testing for expression of the endonuclease when only one ORF is present or only one ORF is expressed. Endonuclease activity will be obtained only when both ORFs are expressed. It may also be possible to reconstitute activity by mixing extracts from two recombinant hosts expressing each ORF separately. The requirement for both ORFs may alternatively be verified by creating deletion or insertion mutations in each of the candidate ORFs separately, and then assessing the endonuclease activity of the resulting recombinant host. For heterodimeric enzymes, both wild-type ORFs will be required for expression of the active endonuclease.

[0039] 2. Cloning the Restriction Endonuclease Gene(s)

[0040] Methods for cloning genes encoding restriction endonucleases are well-established, and are described in numerous publications and patents (Howard et al., Nucleic Acids Res. 14:7939-7951 (1986); Brooks et al., Gene. 74:13 (1988); Lunnen et al., Gene. 74:25-32 (1988); Brooks and Howard, U.S. Pat. No. 5,320,957). These methods will not be repeated here.

[0041] Restriction genes have the capacity to harm cells because the endonucleases they encode can cleave the cell's own DNA. A degree of DNA cleavage can be repaired, but an excess leads to chromosome disintegration and cell death. In order to clone a restriction gene, then, precautions must generally be taken to avoid excessive DNA cleavage. These precautions include maintaining the restriction gene in a non-functional condition; maintaining it in a functional but unexpressed condition; and, protecting the DNA from cleavage by site-specific modification. The latter measure, achieved by cloning into the cell contemporaneously the genes for appropriate modification methyltransferase enzymes, is the preferred one, although under-expression of the endonuclease gene(s), by the use of vectors such as that described in International Publication No. WO 99/11821 can be a useful supplement in cases where complete DNA modification proves difficult to achieve.

[0042] Since the restriction enzymes with which we are concerned, here, recognize asymmetric DNA sequences, two methyltransferases will usually be required to protect the DNA of the host from cleavage, one for modifying each strand of the recognition sequence. Preferably, these methyltransferases will be the ‘cognate’ enzymes that accompany the restriction endonuclease naturally, in the cell from which it derives, but hetero-specific methyltransferases might be used as an alternative (Wilson and Meda, U.S. Pat. No. 5,246,845). The genes for pairs of cognate methyltransferases usually lie next to each other in the genome. In some instances they occur as separate proteins (e.g. the BbvCI system; see Example 1), and in other instances they are fused together (e.g. FokI (Looney et al., Gene. 80:193-208 (1989); Sugisaki et al., J. Biol. Chem. 264:5757-5761 (1989)). The gene(s) for the endonuclease usually lies alongside the genes for its cognate methyltransferases, and the entire set can usually be cloned on one continuous section of DNA. Methyltransferases range in length between 350 aa to 700 aa, and so two methyltransferases will require approximately 2-4 kb of genetic coding capacity. The endonuclease gene(s) requires 1.5-3 kb of coding capacity more, and so the total amount of DNA that needs to be cloned to isolate the complete restriction-modification system is expected to be 3.5-7 kb. Of course, larger and smaller values are also possible.

[0043] Cloning the gene(s) for the cognate methyltransferases is preferred for two reasons: once cloned, they can serve to protect the new host cell from harm when the endonuclease is present; and, because the genes for endonucleases and methyltransferases usually lie next to each other, and the latter can be recognized and selected-for rather easily, methyltransferases ‘flag’ the locations of endonuclease genes within genomes, simplifying their subsequent cloning. Cloning the gene(s) for the methyltransferases, although preferred, might not be essential, however. If the endonuclease is a heterodimer, and neither subunit alone manifests catalytic activity, the genes for the subunits could be cloned individually in the complete absence of DNA modification. In such cases, the active enzyme could be assembled subsequently, in vitro, by mixing together preparations of the individual subunits.

[0044] Nicks in DNA are repaired more readily in cells than are double-strand breaks, and so DNA-nicking enzymes are likely to be less harmful to unmodified cells than are DNA-cleaving enzymes. Consequently, it might be possible to maintain a nicking endonuclease in a functional, expressed, state within an under-modified host cell, and especially if the cell overexpressed a DNA ligase enzyme to repair the nicks. In this event, engineered nicking endonucleases might be maintained and expressed in cells in the complete absence of methyltransferases.

[0045] 3. Identifying the CATALYTIC Sites

[0046] Most restriction endonucleases bind a divalent cation—manganese or magnesium—at each of their catalytic sites. The cation plays an essential role in activating the water molecule which then hydrolyses the phosphodiester section of the DNA backbone to cause its rupture. The structures of the catalytic sites of several restriction endonucleases, determined by x-ray crystallography, establish that these cations are held in place in the enzyme by coordination to the side-chains of two acidic amino acids, either aspartate (D) or glutamate (E). Usually, the coordinating acidic residues are separated in the protein by a loop of variable length—10 to 30 amino acids—and sequence. Proline (P) sometimes precedes the first acidic residue, and lysine (K) usually follows the second acidic residue, 2 or more residues later in the sequence (FIG. 1). The acidic residues, and the lysine are essential for catalysis; substitution of either by uncharged amino acids renders the catalytic sites inert. Based on a few examples, a loose catalytic site motif for restriction endonucleases has been defined as PD . . . 10-25 . . . EXK . . . , where X is hydrophobic.

[0047] Inspection of the amino acid sequences of restriction endonucleases reveals a PD-EXK motif to be present in some of them, hinting that these amino acids might form the catalytic sites in these enzymes. The amino acid sequences of restriction endonucleases vary greatly, however, and because of this variability, it often not possible to identify this motif with any degree of certainty by mere inspection of the sequence. In these cases, a combination of computer programs that predict protein structure, and that ‘thread’ unknown sequences through known structures, can be used to narrow the range of possibilities (on-line ref.). Since amino acids critical to enzyme function are those most resistant to change during evolutionary divergence, global amino acid sequence alignments between related endonucleases can be used to identify conserved regions within which the catalytic site residues are likely to lie. Experimental approaches, based on Fenton chemistry for example, can also be used to roughly locate the catalytic site residues within an endonuclease.

[0048] If the restriction endonuclease is a heterodimer, one set of catalytic site residues will likely be present in one subunit, and a second set will likely be present in the other subunit. If the endonuclease is a monomer, one set of catalytic site residues will likely be present in the amino-terminal half of the protein, and a second set will likely be present in the carboxy-terminal half. It is not necessary to identify both sets of catalytic site residues in order to practice the present invention; one set is sufficient if the strand-specificity of the resulting nicking enzyme is immaterial.

[0049] 4. Catalytic Site Mutagenesis.

[0050] If the likely catalytic site residues within the endonuclease can be identified, one or more of them is targeted for site-specific mutagenesis. Site-specific mutagenesis may be performed using a defined oligonucleotide to create a specific substitution—alanine, for example—or a degenerate oligonucleotide to create a variety of different substitutions. The D, E, and K residues are primary targets for mutagenesis; the neighboring residues are secondary targets. Changes in the former are expected to abolish catalytic activity because these residues are critical to the catalytic mechanism. Changes in the latter are likely to have less predictable, and possibly intermediate effects, because these residues affect the spatial relationships between the catalytic residues rather than the catalytic residues, themselves.

[0051] The goal of the mutagenesis is to inactivate the catalytic site without perturbing the rest of the enzyme. The mutated subunit, or domain, must fold and associate with the other properly; the assembled enzyme must bind to the recognition sequence properly; and the unaltered catalytic site must continue to function properly. Our preferred approach is to perform site-directed mutagenesis using degenerate oligonucleotides that create an array of different mutants, and then to screen among these individually to identify those that behave as we wish.

[0052] Once synthesized, the mutagenic oligonucleotide can be used on its own using the Kunkel mutagenesis procedure, for example, or it can be used together with a reverse primer in a PCR amplification procedure. If the amino acid residues comprising the catalytic sites of the endonuclease cannot be identified unambiguously, but can be narrowed down to a number of possibilities, then mutagenic oligonucleotides can be synthesized to alter each of these possible sites in turn with the expectation that one or other of them will prove to be correct.

[0053] In the event that the catalytic site residues cannot be identified with any confidence, a strategy of non-targeted mutagenesis of the endonuclease gene(s) can be adopted. Alanine scanning of the acidic residues in the protein(s) can be performed in vitro, or ‘dirty’ PCR could be used to create mutations at random. Random mutagenesis could also be performed in vivo by propagating the cloned endonuclease gene in a mutator host such as mutD, or in the presence of chemical mutagens such as 2-aminopurine, or nitrosoguanidine, or during exposure to ultraviolet radiation, or ionizing radiation. Random mutagenesis can be also performed in vitro by reacting the DNA with hydroxylamine, or by amplifying it under conditions that favor base mis-incorporation. Other procedures, such as transposon-mediated insertion mutagenesis, may also be used although they might be less successful because the changes they induce are liable to alter the protein structure, and render the enzyme inactive.

[0054] 5. Screening for Nicking Activity.

[0055] Following mutagenesis, individuals are examined to establish the catalytic activities of the endonucleases they encode. Three outcomes are expected: some enzymes will continue to catalyze DNA cleavage; others will be entirely inactive; and yet others will display the desired nicking activity. The preferred method for distinguishing between these three types is to assay crude preparations of the enzymes on a supercoiled plasmid DNA substrate that contains two or more recognition sites for the enzyme. Plasmid DNA molecules migrate at different rates during electrophoresis depending upon their topologies. The naturally occurring, negatively supercoiled, form migrates more rapidly than the full-length, linear, form, which in turn migrates more rapidly than the nicked, ‘open-circular’ form. These three topological forms separate during high-resolution gel electrophoresis, and the proportion of each in a sample can be assessed from the relative intensities of the corresponding bands.

[0056] Clones expressing mutant enzymes are cultured individually, harvested by centrifugation, and resuspended in a small volume of buffer. The resuspended cultures are sonicated to disrupt the cells, and the insoluble material is removed by centrifugation. Aliquots of the cell extracts are then incubated with supercoiled plasmid DNA in digestion buffer. If the endonuclease is heterodimeric, it is feasible to clone and mutagenize the subunits separately, and to reconstitute the mutant enzymes in vitro, by mixing together the subunits at the time that the assay is performed, or beforehand. Following incubation, the digested plasmid DNA is de-proteinized, and then electrophoresed through an agarose slab gel in the presence of ethidium bromide. Control preparations of undigested, nicked, linear, and cleaved plasmid DNA are electrophoresed in separate lanes alongside the samples to facilitate interpretation of the results. Following electrophoresis, the gel is illuminated with ultraviolet light and photographed.

[0057] Nicking endonucleases are identified by their ability to convert supercoiled DNA to the open-circular form, without the formation of linear molecules or fragments. Once such enzymes are identified, the nicking activity is characterized. The continued sequence-specificity of the enzyme is checked on a supercoiled plasmid DNA substrate that contains no recognition sites. No nicking should be seen; incubation with the enzyme should not alter the relative amounts of the supercoiled, linear, and open-circular forms present naturally in the DNA sample.

[0058] The sequence-specificity and the strand-specificity of the endonuclease is then characterized by analysis of its nicked DNA products. The enzyme is incubated with a DNA molecule containing one or more recognition sites. The DNA is purified and sequenced in both directions across the nick site. When a nick is present in the template strand, the sequencing ladder stops abruptly at the nick. When a nick is present in the complimentary, non-template strand, the sequencing ladder is unaffected. Sequencing across a nick, then, produces a continuous ladder in one direction and a discontinuous ladder in the other direction. The ladder that is discontinuous reveals which strand of the DNA contains the nick, and the position of the discontinuity within the ladder reveals the position of the nick within the sequence. Nicking should occur only at the recognition sequence, and only once per occurrence of the recognition sequence. Only one strand should become nicked, and the position of that nick should be the same as one or other of those catalyzed by parental, cleaving enzyme.

[0059] If only a small number of mutants need to be screened, as would occur if mutagenesis were targeted to just the catalytic site residues, then individual clones can be selected at random and examined in turn. If a large number of clones need to be screened, however, as would occur if mutagenesis were non-targeted, then an initial round of screening in vivo is preferred to identify the rare individuals likely to synthesize a nicking enzyme. In vivo screening might be achieved using an unmodified bacterial host that induced the expression of a reporter gene in response to DNA damage. Provided the endonuclease is well-expressed, cells receiving a normal enzyme will likely die due to excessive DNA cleavage. Cells that receive a nicking enzyme, or an inactive enzyme will likely survive, but only the former will express the reporter gene. The reporter gene can be chosen to render the cells resistant to an antibiotic, or independent of some nutritional requirement, in which cases cells carrying a nicking endonuclease can be isolated selectively. Or, the reporter gene can be chosen to give a chromogenic or fluorescent signal upon expression, in which case cells carrying a candidate nicking endonuclease can be recognized by their coloration. Once potential nicking mutants have been identified in vivo, the catalytic properties of their endonucleases can be examined in vitro.

[0060] The present invention is further illustrated by the following Examples. These Examples are provided to aid in the understanding of the invention and are not construed as a limitation thereof.

[0061] The references cited above and below are herein incorporated by reference.

EXAMPLE 1

Engineered Nicking Derivatives of BbvCI

[0062] 1.0 Overview

[0063] The method for forming a nicking endonuclease starting with restriction endonuclease R.BbvCI is summarized in FIG. 13. Nicking enzyme derivatives of R.BbvCI are obtained by inactivating the active site for cleavage in either subunit of this heterodimeric endonuclease without interfering with the proper subsequent assembly of the enzyme. Appropriate mutations in the enzyme can be created by making mutational changes in amino acids, individually or in combination, that comprise the active site, or that influence its chemistry or organization; and then assessing the nicking activity of enzyme produced by each mutant. The magnitude of this effort may be reduced by focusing on regions conserved in several different but related enzymes. In one preferred embodiment, changes are introduced by the steps of:

[0064] 1.0.1 Identifying a conserved region by alignment of several members of this class of enzymes. Conceptual translations of five genes were employed: the two subunits of BbvCI, termed BbvCI-R1, BbvCI-R2, and three conventional homodimeric Type II endonucleases that recognize related, palindromic, sites: Bsu36I, BlpI, and DdeI. These genes exhibit limited homology in discrete, conserved, blocks. One conserved block contained the sequence EXK. This motif was judged to be the likely active site for cleavage, in which changes may be expected to abolish cleavage but still enable assembly of a conformationally native complex in which the other subunit would still be able to cleave. These were judged favorable sites for mutagenesis.

[0065] 1.0.2 Generating mutations within the favorable region by cassette mutagenesis. This process comprised the steps of:

[0066] a) designing two mutagenic primers for inverse PCR, one for each gene, bbvCIR1 and bbvCIR2. These mutagenic primers were designed such that the nucleotides encoding the EXK motive included 20% random nucleotides, and 80% the correct nucleotide at each of the nine positions. In each mutagenic primer, the region encoding the EXK motif was flanked by the unique sequence of the respective gene;

[0067] b) conducting mutagenic PCR (as disclosed in Molecular Cloning, A Laboratory Manual, Sambrook, 3. and Russel D. W., Cold Spring Harbor Laboratory, pp 8.81-8.95 (2001)) employing in separate reactions i) one mutagenic primer for bbvCIR1 and a unique primer directed in the opposite direction from the mutagenic primer and immediately to its 5′ side; and ii) one mutagenic primer for bbvCIR2 and a unique primer directed in the opposite direction from the mutagenic primer and immediately to its 5′ side, such that the entire plasmid vector was amplified;

[0068] c) ligating the PCR products to form a population circular molecules;

[0069] d) transforming an appropriate host (expressing both methyltransferases) separately with the two mutagenized populations targeting bbvCIR1 and bbvCIR2 to obtain colonies on selective plates; and

[0070] e) for isolated members of each population, testing for cleavage activity in crude extracts, by the steps of:

[0071] i) growing cultures of the candidate colonies;

[0072] ii) centrifuging the cultures to obtain cell pellets;

[0073] iii) resuspending the cultures in lysis buffer;

[0074] iv) lysing the resuspended cultures and clarifying them by centrifugation;

[0075] v) withdrawing aliquots of the clarified extracts to assay tubes containing substrate plasmid DNA and digestion buffer;

[0076] vi) incubating the assay tubes to allow enzyme-induced cleavage to occur; and

[0077] vii) separating the plasmid DNA products by high-resolution gel electrophoresis and assessing whether no cleavage, single-strand cleavage, or double-strand cleavage, has occurred.

[0078] Ideally, the substrate DNA is a plasmid that contains two or more well separated sites for cleavage. Under such circumstances, extracts containing inactive enzyme do not substantially alter the mobility of the various forms of the plasmid. Extracts containing wild-type enzyme abolish the supercoiled, linear, and open-circular forms of the plasmid and produce two (or more) linear fragments in their place. And extracts containing nicking enzyme abolish the supercoiled plasmid form, converting it to open-circular form, without affecting the linear form

[0079] 1.0.3 Testing mutants that appear to nick by alternative procedures to confirm that they have this activity. Such procedures include, but are not limited to, sequencing through nicked sites and sequential nicking with complementary mutants, each defective in the activity of one of the two subunits. Most preferably, candidate enzymes are tested by the first procedure, comprising the steps of:

[0080] a) incubating DNA containing at least one site for cleavage with purified or semi-purified enzyme;

[0081] b) purifying this DNA; and

[0082] c) using it as a substrate for DNA sequencing across the site in both directions.

[0083] Nicking is indicated when the sequence in one direction continues across the site (i.e., the template strand is continuous) while the sequence in the other direction terminates abruptly at the site (i.e., the other strand is interrupted by a nick).

[0084] In the second procedure, extracts of mutants thought to nick different strands are mixed together and the mixture is assayed for double-strand cleavage activity. While neither enzyme alone should catalyze double-strand cleavage, the mixture should be able to do so, either as a result of double-nicking, first on one strand by one enzyme, then on the complementary strand by the other, or by reassociation of the unmutated subunit of each enzyme to produce a fully-wild-type enzyme.

[0085] In this manner mutations in BbvCI-R1 and BbvCI-R2 are identified that enable cleavage of one strand but not the other at BbvCI sites. Two such representative nicking mutants are designated BbvCI #1-37 and BbvCI #2-12.

[0086] 1.1 The Choice of BbvCI

[0087] BbvCI is a Type IIt restriction endonuclease synthesized by the bacterium Bacillus brevis C. It recognizes an asymmetric, seven base-pair nucleotide sequence in DNA, and cleaves symmetrically within this sequence two bases in from each of the 5′ ends:

[0088] 5′-C Cˆ T C A G C-3′

[0089] 3′-G G A G Tˆ C G-5′

[0090] The unusual ability of BbvCI to cleave within a sequence that is asymmetric suggested to us that the subunits, or domains, of this enzyme might differ. Hence BbvCI is a candidate for conversion to a nicking enzyme by the present invention.

[0091] 1.2 Cloning the BbvCI Restriction Endonuclease Genes

[0092] 1.2.1 Vector preparation: To permit the BbvCI genes to be cloned by the ‘methylase-selection’ method, plasmid vector pUC19 was modified prior to library construction by insertion of two BbvCI sites. The first site was inserted between the AatII and EcoO109I sites in the vector, and the second between the two TfiI sites. 1.6 microgram of pUC19 DNA was digested with 20 units each of AatII (NEB Product # R0117, New England Biolabs, Inc., Beverly, Mass.) and EcoO109I (NEB Product # R0503, New England Biolabs, Inc., Beverly, Mass.) in 50 microL of NEBuffer 4 (50 mM potassium acetate, 20 mM Tris acetate, pH 7.9, 10 mM magnesium acetate, 1 mM dithiothreitol) plus 100 microgram/ml BSA for 1 h at 37° C. The following pair of complimentary oligonucleotide linkers were synthesized: 1 bvcpuc-1 (NEB #193-44) 5′-GGCCTCAGC-3′ bvcpuc-2 (NEB #193-44) 5′-GGCGCTGAGGCCACGT-3′ (SEQ ID NO:64)

[0093] The oligonucleotides were suspended in MQ water (autoclave-sterilized, Millipore Q UV-plus, water) to 40 micromolar, and 10 microL of each was added to the restriction enzyme digestion, together with 8 microL of 10×T4 DNA ligase buffer (500 mM Tris-HCl, pH 7.5, 100 mM magnesium chloride, 100 mM dithiothreitol, 10 mM ATP, 250 microgram/ml BSA) and 800 u (2 microL) of T4 DNA ligase (NEB Product # M0202; New England Biolabs, Inc., Beverly, Mass.). After 1 h at room temperature, 1 microL of the reaction was used to transform competent E. coli ER2502. The transformed cells were plated onto L-Broth agar plates containing Ampicillin, and incubated overnight at 37° C. Individual colonies were inoculated into 10 ml L-Broth+Ampicillin, grown overnight, and plasmid DNA was mini-prepped from each. The plasmids were digested with BbvCI (NEB Product # R0601, New England Biolabs, Inc., Beverly, Mass.) to identify those containing the inserted BbvCI site. One of these, PUB-AE, was used as the substrate for inserting the second BbvCI site. PUB-AE was digested with 8 u TfiI (NEB Product # R0546, New England Biolabs, Inc., Beverly, Mass.) for 1 h at 65° C., and the following pair of complimentary oligonucleotide linkers were added, as before, together with 8 microL of T4 DNA ligase buffer and 3 microL of T4 DNA ligase: 2 bvcpuc-3 (NEB #193-63) 5′-AATGGCCCTCAGCC-3′ (SEQ ID NO:65) bvcpuc-4 (NEB #193-64) 5′-ATTGGCTGAGGGCC-3′ (SEQ ID NO:66)

[0094] After incubation at room temperature and subsequent transformation, plasmids containing two BbvCI sites were identified. One of these, designated pUB, was purified by the cesium chloride-ethidium bromide gradient method (Sambrook, Fritsch, and Maniatis, Molecular Cloning, Cold Spring Harbor Laboratory Press, 1989, pp. 1.42-1.46.) and used as the vector for the B. brevis DNA libraries.

[0095] 1.2.2 B. brevis DNA Preparation: 5 g of B. brevis NEB 1030 cell paste was resuspended in 20 ml of 25% sucrose, 50 mM Tris, pH 8.0. 10 ml of 0.25 M EDTA, pH 8.0, was added followed by 60 mg of lysozyme in 6 ml of 0.25 M Tris, pH 8.0. The cells were incubated on ice for 1 hr and then lysed by the addition of 24 ml of 1% Triton X-100, 50 mM Tris pH 8.0, 62.5 mM EDTA, followed by 5 ml of 10% SDS. 70 ml of phenol (equilibrated to pH 8.0) and 60 ml of chloroform was added, and the lyzate was gently emulsified. The emulsion was centrifuged to to separate the phases, and the viscous upper layer containing the DNA was withdrawn and extracted a second time with phenol and chloroform. The centrifuged upper layer was withdrawn and dialyzed for 24 hr at 4° C. against several changes of 10 mM Tris, 1 mM EDTA, pH 8.0. The dialyzed DNA was digested at 37 ° C. for 1 hour with RNase A at a final concentration of 100 microgram/ml, and then the DNA was precipitated by the addition of NaCl to a final concentration of 0.4 M, and 0.55 vol of 100% isopropanol. The precipitated DNA was spooled out of solution on a glass rod, air-dried, and then re-dissolved in 10 mM Tris, 1 mM EDTA, pH 8.0. This procedure yielded 4 ml of high-molecular-weight DNA at a concentration of 200 microgram/ml.

[0096] 1.2.3 B. brevis library preparation: Purified B. brevis DNA was separately digested with AciI (NEB Product # R0551, New England Biolabs, Inc., Beverly, Mass.), BstYI (NEB Product # R0523, New England Biolabs, Inc., Beverly, Mass.), ClaI (NEB Product # R0197 New England Biolabs, Inc., Beverly, Mass.), HpaII (NEB Product # R0171, New England Biolabs, Inc., Beverly, Mass.), and Sau3AI (NEB Product # R0169, New England Biolabs, Inc., Beverly, Mass.). The digested DNAs were ligated with appropriately cleaved pUB vector DNA and used to transform E. coli. Populations of transformants were cultured in L-broth plus ampicillin, and the plasmids they harbored were purified by CsCI-ethidium bromide gradient centrifugation.

[0097] 1.2.4 Selective digestion: 2 microgram of each purified plasmid library was digested with 10 u of BbvCI for 4 h at 37° C. The digestions were transformed into E. coli and plated on LB agar plates containing Ampicillin. Transformant colonies were picked individually and cultured, and their plasmids were mini-prepped for analysis.

[0098] 1.2.5 Clone isolation and analysis: The plasmid minipreps were digested with BbvCI to ascertain their modification status, and with other restriction enzymes to check for common fragments among the inserts. One BbvCI-modified isolate, containing a 3.0 kb ClaI-fragment, was sequenced and found to contain two complete open reading frames, and the beginning of a third. Amino acid sequence analysis of the complete ORFs showed each to code for a 5-cytosine methyltransferase, indicating that these ORFs were the two BbvCI methyltransferase genes. Analysis of the genomic DNA downstream of the incomplete ORF was performed by inverse PCR. Genomic DNA was digested with BsrGI (NEB Product # R0575, New England Biolabs, Inc., Beverly, Mass.), BstYI, HindIII (NEB Product # R0104, New England Biolabs, Inc., Beverly, Mass.), and NciI (NEB Product # R0196, New England Biolabs, Inc., Beverly, Mass.), self-ligated, and then amplified using oligonucleotide primers complementary to DNA sequence within the known region. The PCR products were sequenced directly. 1.6 kb of additional sequence was obtained in this way, completing the first ORF and revealing the presence of a second complete ORF downstream from the first and bearing substantial amino acid sequences homology to it. This suggested that the BbvCI endonuclease was a heterodimer, and that the ORFs were the genes for its two subunits. The nucleotide sequence determined for the BbvCI endonuclease and methyltransferase genes is shown in FIG. 1. The predicted amino acid sequences of the endonuclease and methyltransferase proteins are shown in FIG. 2.

[0099] Purified BbvCI endonuclease, analyzed by SDS-PAGE, was found to migrate as two distinct bands of approximately 31 kDa and 32 kDa, the sizes predicted for the products of the ORFs. Deletion analysis demonstrated that the 31 kDa band was the 275-aa subunit encoded by the first, R1, ORF, and that the 32 kDa band was the 285-aa subunit encoded by the second, R2, ORF (FIG. 14). Both ORFs were found to be essential for the expression of endonuclease activity. The amino acid sequences of first few residues of the two subunits were found to match those predicted from the nucleotide sequences of the two ORFs, confirming their identities. The N-terminal sequence of the 31-kDa band was found to be MINEDFFIY . . . , corresponding to the first 9 codons (SEQ ID NO: 4) of the R1 gene, and that of the upper, 32-kDa band was found to be MFNQFNPLVY . . . , corresponding to the first 10 codons (SEQ ID NO: 5) of the R2 gene.

[0100] 1.2.6 Expression of the BbvCI endonuclease: To 5 construct a stable expression system for the BbvCI endonuclease, the genes for the BbvCI methyltransferases were first PCR-amplified using the following primers, and ligated into the SacI (NEB Product # R0156, New England Biolabs, Inc., Beverly, Mass.) and SphI (NEB Product # R0182, New England Biolabs, Inc., Beverly, Mass.) sites in the polylinker of the plasmid vector pIH919. pIH919 consists of the LacZ region and polylinker from pUC18, excised with PvuII (NEB Product # R0151, New England Biolabs, Inc., Beverly, Mass.) and HindIII, and ligated into pACYC184 digested with NruI (NEB Product # R0192, New England Biolabs, Inc., Beverly, Mass.) and HindIII. 3 M1-BbvCI fwd (NEB #205-68) 5′-GGGGCGAGCTCAGGAGGTTAAAATATGGAATCT (SEQ ID NO:67) GAAACACGTCAAAATATA-3′ M2-BbvCI rev (NEB #205-69) 5′-GCAATAGCATGCTTACTACCCCCATAATTGTAG (SEQ ID NO:68) TTGTTCC-3′

[0101] The resulting plasmid was transformed into E. coli ER2688 to create a host cell protectively modified against digestion by the BbvCI endonuclease. The genes for the two BbvCI endonuclease subunits were then PCR-amplified using the following primers, ligated into the SacI and SphI sites of pUC19, and transformed into the protected host. 4 R1-BbvCI fwd (NEB #205-72) 5′-GGGGGCATGCTAAGGAGGTTTAAAATATGATTA (SEQ ID NO:69) ACGAGGACTTTTTTATTTAT-3′ R2-BbvCI fwd (NEB #205-71) 5′-GCAATAGAGCTCTTATTATGGTCGATTAAACAC (SEQ ID NO:70) TTTACCGCG-3′

[0102] The resulting strain, E. coli ER2688 containing the bbvCIM1 and M2 genes in pACYC184, and the bbvCIR1 and R2 genes in pUC19, produced approximately 5 million units of R.BbvCI endonuclease per gram of cells.

[0103] 1.3 Identifying the CATALYTIC Sites in BbvCI

[0104] 1.3.1 Analysis of related endonuclease, BlpI and Bsu36I: To assist us in identify the catalytic site residues in the BbvCI endonuclease subunits, two additional restriction endonucleases were cloned and sequenced. Bsu36I and BlpI, homodimeric Type II endonucleases, recognize the symmetric sequences CCˆ TNAGG, and GCˆ TNAGC, respectively. The Bsu36I half-site (CCˆ Tn) resembles the left half-site of BbvCI (CCˆ Tc), and the BlpI half-site (GCˆ Tn) resembles the right half-site of BbvCI (GCˆ Tg). These similarities suggested that all three enzymes might be evolutionarily related, and if this were the case their catalytic site residues were expected to reside within regions common to all of them.

[0105] Bacillus lentus (NEB #819) DNA was purified in the manner described for B. brevis (see section 1.2.2, above), and digested to completion with HindIII. The HindIII-digested DNA was ligated at 17° C. overnight into a HindIII-cleaved, CIP-treated, pUC19 plasmid vector derivative into which a single BlpI site had been introduced beforehand in the manner described for BbvCI (see section 1.2.1, above). The ligated DNA was used to transform competent E. coli ER2426 cells, and the transformed cells were propagated in culture in the presence of ampicillin. Plasmid DNA was prepared from the culture and digested at 37° C. for 2 hr with BlpI (NEB Product # R0585, New England Biolabs, Inc., Beverly, Mass.) in NEBuffer 4, and then dephosphorylated at 37° C. for 1 hr with calf intestinal phosphatase (NEB Product # M0290, New England Biolabs, Inc., Beverly, Mass.). The DNA was transformed back into competent ER2426 cells, and plasmid DNA was isolated from individual transformants and checked for resistance to digestion by BlpI. One resistant plasmid, isolate #20, expressed BlpI endonuclease activity and contained a single 3.6-kb HindIII-fragment insert. This fragment was sequenced and found to contain the complete BlpI endonuclease gene and the complete BlpI methyltransferase gene. The nucleotide sequence of the BlpI genes is shown in FIG. 3, and the predicted amino acid sequences of the proteins, in FIG. 4.

[0106] Bacillus subtilis 36 (NEB #440) DNA was purified in the manner described for B. brevis, and digested partially with HindIII. The HindIII-digested DNA was ligated at 17° C. overnight into a HindIII-cleaved, CIP-treated, pUC19 plasmid vector derivative into which two Bsu36I sites had been introduced beforehand in the manner described for BbvCI. The ligated DNA was used to transform competent E. coli ER2683 cells, and the transformed cells were propagated in culture in the presence of ampicillin. Plasmid DNA was prepared from the culture and digested at 37° C. overnight with Bsu36I (NEB Product # R0524, New England Biolabs, Inc., Beverly, Mass.) in NEBuffer 3 (100 mM NaCl, 50 mM Tris-HCl, pH 7.9, 10 mM magnesium chloride, 1 mM dithiothreitol) supplemented with 100 microgram/ml BSA. The digested DNA was treated with exonucleases III (NEB Product # M0206, New England Biolabs, Inc., Beverly, Mass.) at 37° C. for 30 min, and then it was transformed back into competent ER2683 cells. Plasmid DNA was isolated from individual transformants and checked for resistance to digestion by Bsu36I. One resistant plasmid, isolate #15, expressed Bsu36I endonuclease activity, and contained 3 HindIII-fragments totaling 4.6 kb. These fragments were sequenced and one, also 3.6-kb in length, was found to contain the complete Bsu36I endonuclease gene and the complete Bsu36I methyltransferase gene. The nucleotide sequence of the Bsu36I genes is shown in FIG. 5, and the predicted amino acid sequences of the proteins, in FIG. 6.

[0107] 1.3.2 Multiple sequence alignment: The amino acid sequences of the BbvCI endonuclease subunits were aligned with those of BlpI, Bsu36I and a third related Type II endonuclease sequenced earlier, DdeI (Sznyter et al., Nucleic Acids Res. 15: 8249-8266 (1987)). Initial alignment was performed using the GCG programs ‘Gap’, and then ‘Pretty’. Final alignment was performed by eye (FIG. 7). Several blocks of sequence conservation were revealed by the alignment. The fourth block contained the tri-peptide sequence EXK (X=C or V). This motif commonly forms part of the catalytic site of restriction endonucleases, and it was judged likely to form part of the catalytic sites in the BbvCI subunits, too.

[0108] 1.4 Mutagenesis of the BbvCI Catalytic Sites

[0109] 1.4.1 Mutagenic PCR primers: The nucleotide sequences of the sections of the BbvCI genes coding for the putative catalytic site EXK residues was consulted. Based on these sequences, the following four 5′-phosphorylated PCR oligonucleotide primers were synthesized. 5 R1‘mutagenic’ (NEB #249-180) 5′-pTTATTAGTTTAGCCGTCGCATGTgaggtcaag (SEQ ID NO:71) ACAAATATCGATAAAAACAAACTTAATGGGTTAGAC -3′ R1 reverse (NEB #249-181) 5′-pCTCTATGACTCTCACTTGCAATTTTAACGTCT (SEQ ID NO:72) ACC-3′ R2‘mutagenic’ (NEB #249-174) 5′-pTGGTTCCTGCTGTAGCGATTgagtgcaagCGG (SEQ ID NO:73) TACCTTGAACGAAACATGCTAGATGAATGTGCTGG- 3′ R2 reverse (NEB #249-175) 5′-pCAATGTTGGTTGTTTCTCCTTCCCCGCCATGT (SEQ ID NO:74) CC-3′

[0110] The two mutagenic, forward, primers overlapped the codons for the putative catalytic site residues, and matched the sense-strand of the BbvCI R genes. The two non-mutagenic, reverse, primers matched the anti-sense strand immediately preceding the mutagenic primers. The 5′-ends of the forward and reverse primer pairs butted up flush with one another, one priming in one direction, the other priming in the opposite direction.

[0111] In the reverse primers, the bases were homogeneous at each position. In the mutagenic primers, the bases at the ends were homogeneous, but the middle 9 bases comprising the 3 codons of the putative catalytic site EVK and ECK (shown in lowercase, above) were heterogeneous. The first and second positions in these three codons were mixtures of all four bases, 80% being the wild-type base shown in lower case, and the remaining 20% being an equal mix of the other three bases. The third position in these codons was a mixture of only two bases, G and C, 80% being the wild-type base, and 20% being the other base. The 80:20 ratio was chosen to generate mutants approximately 40% of which would incorporate one amino acid change, 40% would incorporate two changes, and 10% would incorporate three. The following binomial equation describes approximately how the likelihood, P, of r mutations occurring within N codons varies with the proportion, p, of mutagenic bases in the mixtures:

P(r)=(N!/(N−r)!r!)×((1−p)3)N−r×(1−(1−p)3)r

[0112] 1.4.2 PCR mutagenesis: Two sets of PCR reactions were prepared, one to mutagenize each subunit. Each reaction was set up as follows: 6 10X ThermoPol Buffer:  150 microL 20X mutagenic primer:  75 microL 20X reverse primer:  75 microL 25X stock dNTP mix:  100 microL 400 ng template DNA:   2 microL MQ water: 1058 microL (Vent ® DNA polymerase:  20 microL) (Deep Vent ® DNA polymerase:  20 microL)

[0113] Notes: 10×ThermoPol buffer is: 200 mM Tris-HCl, pH 8.8, 100 mM KCl, 100 mM ammonium sulfate, 20 mM magnesium sulfate, 1% Triton X-100.

[0114] 20×primer is 6 micromolar.

[0115] 25×stock dNTP mix is 5 mM each of dATP, dCTP, dGTP, and dTTP.

[0116] Template DNA was purified plasmid pUC19(bbvCI R1,R2) at 200 microgram/ml.

[0117] The reactions were prepared on ice. The two DNA polymerases were added last, and immediately after wards the reactions were dispensed 100 microL at a time into 15 microtubes pre-warmed to 95° C. The reactions were cycled as follows: 95° C., 45 sec (denature); 4° C., 1 min (anneal); 72° C., 4 min (polymerize)×30 cycles. 72° C., 8 min (completion); 40° C. (hold).

[0118] 1.4.3 Mutant isolation: The two sets of PCR reactions were separately pooled and electrophoresed in 1% agarose gels. The prominent band corresponding to the full-length PCR product from each reaction was sliced from the gel and the fragment was recovered, phenol-chloroform extracted, precipitated, and resuspended in MQ water to a concentration of approximately 100 microgram/ml. The purified fragments were separately self-ligated as follows: 7 PCR fragment @ 100 microgram/ml: 20 microL 10X ligation buffer:  6 microL MQ water: 32 microL conc. T4 DNA ligase (2G u/ml)  2 microL

[0119] The ligations proceeded at room temperature for 3.5 hr., and then they were terminated by the addition of chloroform. An aliquot of each was analyzed by agarose gel electrophoresis and ligation was found to be extensive in both reactions. 1 microL aliquots of the ligations were used to transform competent E. coli ER2502. The transformations were plated onto LB+Ampicillin plates and incubated overnight at 37° C. 60 colonies from each set were inoculated into 10 ml L-Broth plus Ampicillin and grown overnight. 1.5 ml of each culture was used for plasmid minipreparation, and 1.7 ml was used to prepare extracts for enzyme assay.

[0120] 1.4.4 Sequence analysis of the mutants: Four oligonucleotide primers were synthesized to sequence in both directions across the mutated regions in the two BbvCI subunit genes: 8 R1(+) (NEB #249-176) 5′-GGTTAAAACCTATTTTCTCTACGAG-3′ (SEQ ID NO:75) R1(−) (NEB #249-177) 5′-CTTGTTTTCGAAGAACATAAATTTC-3′ (SEQ ID NO:76) R2(+) (NEB #249-178) 5′-ATTTAAGTCTCAGCTTTAATCCAC-3′ (SEQ ID NO:77) R2(−) (NEB #249-179) 5′-GAGTTCCGCTGGTGCCGAAGTATG-3′ (SEQ ID NO:78)

[0121] Isolates from the PCR reaction using the R1 mutagenic primer were sequenced with the R1(+) and R1(−) sequencing primers. Isolates from the reaction using the R2 mutagenic primer were sequenced with the R2(+) and R2(−) primers. The resulting sequences were examined to identify isolates that contained changes other than those affecting the three catalytic site amino acids. Erroneous isolates so-identified, containing mutations elsewhere, were discarded. The sequences of the remaining, genetically intact, isolates were then examined to determine the amino acids present at their catalytic sites. Representative results are shown in FIGS. 17 and 18.

[0122] 1.5 Nicking Activity Assays

[0123] 1.5.1 Crude extract assays: 1.7 ml of overnight cell culture of each mutant was microcentrifuged, the supernatant was discarded, and the pellet was frozen at −20° C. for up to one week. When convenient, the pellet was thawed on ice and resuspended in sonication buffer (100 mM NaCl, 20 mM Tris, pH 8.0, 1 mM EDTA, 5% glycerol) plus 10 mg/ml lysozyme. After 30 min on ice the suspension was sonicated with a micro tip for 10 sec at 60% duty using a Heat Systems-Ultrasonics Sonicator at power setting 5. After microcentrifugation to remove cell debris, 1 microL of the supernatant was incubated with 1 microgram of substrate DNA in 50 microL of NEBuffer 4 for 1 h at 37° C. The reactions were de-proteinized by the addition of 10 microL 10×stop dye containing 0.5% SDS, or 10 microL 10×stop dye followed by 5 microL phenol and 5 microL chloroform. The reactions were emulsified, microcentrifuged to clarify, and then the DNA digests were electrophoresed through 1% agarose gels for analysis.

[0124] Two different DNA substrates were used for characterizations: phage lambda DNA, and supercoiled pUB plasmid DNA. Lambda enabled easy assessment of DNA-cleaving activity, and pUB, easy assessment of DNA-nicking activity. Some isolates produced wild type enzymes that cleaved both DNA substrates (FIG. 8). The mutations in these did not affect the catalytic activities of the subunits. Other isolates produced inactive enzymes that neither cleaved nor nicked the DNA substrates. The mutations in these, we presume, disrupted the subunits so severely that all catalytic activity was abolished. And yet other isolates produced nicking enzymes that converted the supercoiled plasmid DNA to the open circular form, and distinctively changed the mobility of the lambda DNA, but generated no full-length linear plasmid fragments, or BbvCI-cleavage fragments. The mutations in these inactivated the catalytic sites in a localised fashion without affecting the rest of the enzyme. The latter mutants were the nicking endonucleases we sought.

[0125] 1.5.2 Nicking endonuclease purification: Enzyme mutants 1-37 (R1E167G:R2+) and 2-12 (R1+:R2E177G) were selected as being representative of nicking mutants with inactive catalytic sites in the R1 subunit, and the R2 subunit, respectively. These endonucleases were purified to homogeneity. 100 liter cell cultures were grown to saturation at 37° C. in L-Broth (10 gm/L Tryptone, 5 gm/L yeast extract, 0.17 M NaCl, 5.6 mM dextrose, 5 mM MgCl2), harvested by centrifugation, and stored frozen at −70° C. 100 gm cell paste was thawed at room temperature and resuspended in 350 ml of column buffer (20 mM Tris-HCl, pH 8.0, 6 mM Beta-mercaptoethanol, 0.25 mM EDTA, 5% glycerol) containing 100 mM NaCl. 500 mg of lysozyme was added, and the suspension was stirred for 30 min at 4° C. The suspension was sonicated for 2×4′ intervals using a Branson Sonifier at 70% duty, and a power setting of 8. The sonicated suspension were centrifuged in 2×250 ml bottles at 13,000 rpm for 30 min in a JA-14 (Beckman) rotor. The supernatant was decanted and recentrifuged twice more, until no cell debris remained.

[0126] The cleared supernatant was loaded at 4 ml/min onto 100 ml of Heparin Hyper-D resin (BioSepra, Inc.) in a XK26 (Pharmacia) FPLC column. The column was washed with 2 column vol of column buffer containing 100 mM NaCl, and then eluted with column buffer containing an increasing NaCl gradient. 5 ml fractions were collected and selected fractions were assayed for enzyme activity in the following manner. 2 microL of each fraction was diluted in 100 microL of NEBuffer 4, and 2 microL of the dilution was added to 25 microL of NEBuffer 4 containing 0.5 microgram of supercoiled pUB DNA or phage lambda DNA. The reactions were incubated at 37° C. for 2 min, stopped and de-proteinized by the addition of 5 microL of 10× stop dye containing 0.5% SDS, and then electrophoresed on an 0.8% agarose gel and visualized. Peak activity eluted from the Heparin column around 0.5 M NaCl.

[0127] Active fractions were pooled to a volume of approximately 100 ml and dialyzed overnight against column buffer containing 50 mM NaCl. The precipitate that formed was removed by centrifugation without loss of enzyme activity. The clarified dialyzate was passed through 24 ml of Source 15S resin (Pharmacia) in a HR16/10 FPLC column. The column was washed with 2 column vols. of column buffer containing 50 mM NaCl. Enzyme activity remained in the flow-through which was brought up to 0.15 M NaCl and passed through 8 ml of Source 15Q resin (Pharmacia) in a HR10/10 FPLC column. The column was washed with 1 column vol of column buffer containing 150 mM NaCl, and enzyme activity again remained in the flow-through. The flow through was loaded directly onto 60 ml of Heparin Hyper-D resin in a XK26 column once more, and eluted with increasing [NaCl] gradient in column buffer. Active fractions peaked at 0.5 M NaCl and were pooled, dialyzed into storage buffer (50 mM NaCl, 20 mM Tris, pH 7.4, 1 mM DTT, 0.1 mM EDTA, 50% glycerol) and stored at −20° C. for short-term storage, or −70° C. for the longer term.

[0128] Enzymes prepared in this way had a final protein concentration of approximately 40 mg/ml. The yields and activities of these enzyme preps are shown in FIG. 16. Analyzed by SDS-PAGE, the enzymes were found to be homogeneous and to migrate as two bands of the expected sizes, and in the approximate ratio of 1:1 (FIG. 9).

[0129] 1.5.3 Characterization of nicking endonuclease activity: Aliquots from the nicking enzyme preps were titrated on supercoiled pUB DNA (two BbvCI sites). Control titrations on the same DNA using EcoRI (1 site; NEB Product # R0101, New England Biolabs, Inc., Beverly, Mass.) and the wild type BbvCI cleaving enzyme were run alongside. The digests were analyzed by high-resolution agarose gel electrophoresis. The product of the nicking enzymes migrated at a slightly slower rate than the full-length linear product of the EcoRI-digestion, confirming that it was not linear. No BbvCI-fragments were produced in the nicking enzyme digests confirming that these enzymes were unable to hydrolyze both strands of the DNA (FIG. 10).

[0130] (Note: Subsequent digestions with the nicking enzyme 1-37 did on occasion produce BbvCI-cleavage fragments at a very low level. This was traced to the presence of variable amounts of manganese in the water. The altered catalytic site of 1-37 (EVK->GVK) is inactive in the presence of magnesium but partially active in the presence of manganese. Other mutants in the R1 subunit do not behave in this way, nor does the equivalent mutation in the catalytic site of the R2 subunit: mutant 2-12 (ECK->GCK) nicks DNA in the presence of either magnesium and manganese.)

[0131] The sequence-specificities of the purified 1-37 and 2-12 endonucleases were tested by incubating them with supercoiled pUC19 DNA, a plasmid containing no BbvCI sites. No nicking by either enzyme was observed. In contrast, in control digestions run alongside, the enzymes nicked pUB DNA to completion, and R.EcoRI but not R.BbvCI, digested pUC19 DNA to completion.

[0132] The strand-specificities of the purified 1-37 and 2-12 endonucleases was tested by incubating supercoiled pUB DNA first with one nicking endonuclease and then with the other. The DNA was cleaned on ‘Zymoclean’ spin columns to remove the first enzyme prior to addition of the second. Incubating with either enzyme nicked the DNA. Incubating with the same enzyme in the second round as was used in the first had no effect, but incubating with the other enzyme resulted in complete DNA cleavage, and the production of fragments identical to those produced by R.BbvCI. This occurred regardless of the order in which the nicking enzymes were added, and it demonstrated that 1-37 and 2-12 nick DNA on opposite strands (FIG. 11).

[0133] The strand-specificities of the purified 1-37 and 2-12 endonucleases was further examined by sequencing in both directions across the nicked BbvCI sites in pUB DNA. pUB DNA was separately nicked with each enzyme. The nicked DNA samples were purified on Zymoclean spin-columns, resuspended in MQ H2O, and then used as substrates for DNA sequencing reactions in both directions across the two BbvCI recognition sites using the following custom-synthesized oligonucleotide primers: 9 Fwd-1 (NEB #254-257) 5′-CACAATTCCACACAACATACGAGC-3′ (SEQ ID NO:79) Rev-1 (NEB #254-258) 5′-TCGCCACCTCTGACTTGAGCGTCG-3′ (SEQ ID NO:80) Fwd-2 (NEB #254-259) 5′-TTCCTTTTTCAATATTATTGAAGC-3′ (SEQ ID NO:81) Rev-2 (NEB #254-260) 5′-CGACACCCGCCAACACCCGCTGAC-3′ (SEQ ID NO:82)

[0134] On DNA nicked with 1-37, the reactions with primers Fwd-1 and Fwd-2 produced continuous sequencing traces through the BbvCI sites, indicating that the ‘bottom’ strand in the recognition sequences—the template strand in these reactions—was intact. In contrast, the reactions using primers Rev-1 an Rev-2 produced taces that stopped abruptly at the BbvCI sites, indicating that the ‘top’ strand in both of the recognition sequences was nicked (FIG. 12). In these latter reactions, the traces ended with . . . GCTGAr, indicting that the template strand was nicked after the second base in the sequence . . . CCˆ TCAGC . . . , at the position indicated by ‘ˆ ’. (The weak, final mixed a and g signal, ‘r’, was an artifact caused by the template-independent addition of purines by the Taq DNA polymerase used for sequencing.) This is the same position at which the wild-type BbvCI endonuclease cleaves this strand.

[0135] Conversely, on DNA nicked with 2-12, the reactions with Fwd-1 and Fwd-2 primers stopped abruptly at the BbvCI sites, and the reactions using the Rev-1 an Rev-2 primers produced continuous ladders (FIG. 12). The traces in the terminated reactions ended with . . . CCTCAr, indicting that the nick on the template strand also occurred after the second base in the recognition sequence GCˆ TGAGG, again the same position at which wild-type R.BbvCI hydrolyzes this strand.

[0136] These results demonstrate that the R1 subunit of BbvCI hydrolyzes the bottom strand of the recognition sequence, and the R2 subunit hydrolyzes the top strand. The wild-type and nicking endonucleases thus have the following specificities: 10 R.BbvCI (R1+: R2+) C C{circumflex over ( )}T C A G C Wt heterodimer G G A G T{circumflex over ( )}C G Nt.BbvCI (e.g. 1-37: R1−: R2+) C C{circumflex over ( )}T C A G C Top-strand nicking enzyme G G A G T C G Nb.BbvCI (e.g. 2-12: R1+: R2−) C C T C A G C Bottom-strand nicking enzyme G G A G T{circumflex over ( )}C G

[0137] 1.5.4 Cation-dependent enzyme activity: The endonucleolytic activities of crude extracts of the nicking mutants were tested in digestion buffer containing (a) only magnesium as the divalent cation; (b) only manganese as the divalent cation; or (c) a mixture of both magnesium and manganese. Most of the mutant enzymes that nicked when only magnesium was present continued to nick, but to lesser extent, when only manganese was present or when both cations were present. However, three enzymes that nicked when only magnesium was present, 1-29 (R1E167D:R2+), 1-37 (R1E167G:R2+), and 2-56 (R1+:R2E177D), displayed substantial cleavage activity in the presence of only manganese, or in the presence of both magnesium and manganese. Evidently, the mutated catalytic sites in these three enzymes are inactive when the cation is magnesium, but partially active when the cation is manganese. The addition of manganese to these three enzymes, then, converts them from DNA-nicking enzymes to DNA-cleaving enzymes.

[0138] The endonucleolytic activities of crude extracts of the nicking mutants were further tested in digestion buffer containing both magnesium and calcium. The presence of calcium at a level as low as 4 mM completely inhibited nicking activity in all of the mutants except 2-56. Mutant 2-56 (R1+:R2E177D) was fully active in 4 mM calcium, but was completely inhibited by 8-10 mM calcium.

EXAMPLE 2

Potential Nicking Derivatives of BstNBI and BbvI

[0139] Type IIs and IIf restriction endonucleases are thought to catalyze DNA cleavage, in general, by transient homodimerization. Just as the occasional Type II endonuclease acts as a heterodimer, however, so too might the occasional Type IIs or IIf endonuclease. This Example is based on just such a prospect. It refers particularly to BbvI and BstNBI, Type IIs endonucleases for which circumstantial evidence of a possible heterodimeric nature exists. If either enzyme indeed acts as a heterodimer, then it is a candidate for conversion to a nicking endonuclease by the present invention. Of the few Type IIf endonucleases discovered to date, none appears likely to be heterodimeric. Such enzymes probably do exist in nature, however, and when they are discovered they will be candidates for conversion to nicking endonucleases by the present invention.

[0140] Heterodimeric Type IIs restriction endonucleases are expected to comprise one large subunit and one small subunit. The large subunit, like a normal Type IIs endonuclease, will comprise two domains, one for sequence-specific DNA-binding, the other for catalysis and dimerization. The small subunit will comprise the second catalytic site and dimerization surface. Dimerization will take place, not between two molecules of the large subunit as occurs in enzymes that homodimerize, but between one molecule of the large subunit and one of the small subunit. The dimerization surfaces on the subunits of the heterodimer will be such that each subunit can associate with the other, but not with itself.

[0141] If we denote sequence-specific DNA-binding as ‘B’, DNA-cleavage as ‘C’, and dimerization as ‘D’, then the normal Type IIs monomer can be described as B+C+D+, and the catalytically active, transient homodimer as B+C+D+:B+C+D+, where ‘+’ signifies wild-type functionality, and ‘−’ (see later) signifies non-functionality. Using the same notation for heterodimeric Type IIs endonucleases, the large subunit can be described as B+C1+D1+, the small subunit as C2+D2+, and the catalytically active heterodimer as B+C1+D1+:C2+D2+. The distinction between the homodimeric enzyme and the heterodimeric enzyme is that in the former, the catalytic sites, C, are identical; the dimerization surfaces, D, are identical and self-compatible; and sequence-specific binding by only one of the two DNA-binding domains, B, is sufficient for strand-hydrolysis to proceed. In contrast, in the heterodimer the catalytic sites, C1 and C2, are different; the dimerization surfaces, D1 and D2, are different and are compatible, but not self-compatible; and there exists only one binding domain by which sequence-specific DNA-binding can occur.

[0142] Predictable differences distinguish heterodimeric Type IIs endonucleases from their homodimeric counterparts. Whereas homodimeric endonucleases are encoded by single genes, heterodimeric endonucleases will be encoded by two genes, one for each subunit. The two genes are likely to lie side-by-side in the genome, in the same orientation. Amino acid sequence analysis of the subunits might reveal the presence of putative catalytic site residues in each. The two subunits might share amino acid sequence homology with each other, and with other restriction endonucleases, particularly in the catalytic site regions. In vivo, the gene for neither subunit alone would manifest DNA-cleavage activity, but together they would. The gene for the large subunit on its own might manifest DNA-nicking activity, or it might be inactive. The gene for the small subunit on its own might manifest non-specific endonuclease activity, or more likely it would be inactive. Mixing the two subunits together in vitro should produce DNA-cleavage activity. If the large subunit alone displayed DNA-nicking activity in vitro, adding the small subunit should convert this nicking activity to DNA-cleaving activity.

[0143] The activity of homodimerizing Type IIs endonucleases is higher on DNA substrates that contain several recognition sites than it is on DNA substrates that contain only one. This should not be true for heterodimeric Type IIs endonucleases; they should display the same activity regardless of the number of recognition sites in the DNA. The same should apply to ‘functionally heterodimeric’ endonucleases whose two ancestral subunits have fused to form a single, monomeric, protein. This ‘cis-stimulation’ effect could be a useful experimental test to rapidly distinguish between homodimeric and heterodimeric/monomeric endonucleases.

[0144] Because the subunits of heterodimers will associate and dissociate at some rate, they are liable to become separated during protein purification. A heterodimeric Type IIs endonuclease would begin with a 1:1 subunit stoichiometry, but during purification one subunit—probably the smallest—could gradually be lost, and the other subunit could come to predominate in the active fractions. The DNA-cleavage activity of such fractions would depend upon how much of the minor subunit remained, and so the overall amount of enzyme activity would decline during purification, giving the enzyme the appearance of being unstable. If the large subunit on its own displayed nicking activity, and the small subunit were lost during purification, active fractions of the enzyme would exhibit an increasing specific activity for nicking as opposed to cleaving, and ultimately the enzyme might appear to be natural nicking endonuclease, and not a cleaving endonucleases at all.

[0145] A small unidentified open reading frame lies next to the ‘endonuclease’ genes in the BbvI and the BstNBI restriction-modification systems (Higgins, et al., Nucleic Acids Res. 29:2492-2501 (2001)). The BbvI ORF potentially encodes a protein of around 150 aa that displays homology to the C-terminal halves of two Type II endonucleases, gb:U13922, and gb:U43736. The BstNBI ORF potentially encodes a protein of approximately 180 amino acids that displays homology to the C-terminal catalytic domains of several Type IIs endonucleases, including that of the N.BstNBI nicking endonuclease encoded by the adjacent gene. The acidic catalytic site residues present in the nicking endonuclease are also present in the ORF. Both the BbvI and the BstNBI ORFs are preceded by substantial ribosome-binding sites, and both are followed by substantial stem-loop transcription terminators that likely isolate them from the adjacent, opposed, methyltransferase genes. Taken together, these features suggest that the ORFs are expressed, and that their products are important components in the proper biological functioning of the BbvI and BstNBI restriction-modification systems. Plausibly, the ORFs could be the genes for the small catalytic subunits of hitherto unrecognized heterodimeric endonucleases, the large subunits of which are encoded by the adjacent ‘endonuclease’ genes. Additional, perhaps similar, systems can be found in GenBank, for example genes BH4001-BH4004 of AP001520 from the Bacillus halodurans genome (Takami, H., et al. Nucleic Acids Res. 28:4317-4331 (2000)). If these endonucleases are indeed heterodimeric, then their two catalytic sites can be independently altered in the ways described in Example 1.

[0146] Although natural R.BstNBI has not yet been discovered, we speculate that co-expressing the bstNBIN gene and the adjacent ORF might produce this enzyme in vivo, and that mixing N.BstNBI with the purified product of the BstNBI ORF might produce it in vitro, too. N.BstNBI nicks the top strand of DNA four bases outside the recognition sequence (Higgins, et al. Nucleic Acids Res. 29:2492-2501 (2001)). If N.BstNBI is the large subunit of a heterodimer, it should be possible to convert the heterodimer into a nicking enzyme of the opposite strand-specificity by inactivating the catalytic site in just the large subunit. The following enzyme activities is thus predicted as follows: 11 N.BstNBI (B+C1+D1+): GAGTCNNNNˆ N . . . natural monomer CTCAGNNNNNN . . . R.BstNBI (B+C1+D1+:C2+D2+) GAGTCNNNNˆ N . . . hypothetical heterodimer CTCAGNNNNˆ N . . . Nt.BstNBI (B+C1+D1+:C2−D2+) GAGTCNNNNˆ N . . . top-strand nicker CTCAGNNNNNN . . . Nb.BStNBI (B+C1−D1+:C2+D2+) GAGTCNNNNNN . . . bottom-strand nicker CTCAGNNNNˆ N . . .

[0147] (In the above sequences the location of the nick on the bottom stand is four bases beyond the recognition sequence.) We hypothesize that R.BbvI might also be naturally heterodimeric, but that the isolated large subunit of this enzyme, unlike that of BstNBI, does not display nicking activity. Consequently, R.BbvCI may be convertible into a nicking endonuclease of either strand specificity to create the following: 12 R.BbvI (B+C1+D1+:C2+D+) GCAGCNNNNNNNNˆ NNNNN . . . hypothetical heterodimer CGTCGNNNNNNNNNNNNˆ N . . . Nt.BbvI (B+C1+D1+:C2−D2+) GCAGCNNNNNNNNˆ NNNNN . . . top-strand nicker CGTCGNNNNNNNNNNNNNN . . . Nb.BbvI (B+C1−D1+:C2+D2+) GCAGCNNNNNNNNNNNNNN . . . bottom-strand nicker CGTCGNNNNNNNNNNNNˆ N . . .

[0148] (In the above sequences, the large subunit is shown to hydrolyze the top strand, and the small subunit, the bottom strand.)

EXAMPLE 3

Potential Nicking Derivatives of FokI

[0149] Dimerization of Type IIs and Type IIf restriction endonucleases is thought to occur, in general, between the catalytic domains of identical enzyme molecules, one bound specifically to the recognition site in DNA, the other perhaps free in solution, or bound to DNA non-specifically (Wah et al., Nature. 388:97-100 (1997); Wah et al., Proc. Natl. Acad. Sci. USA 95:10564-10569 (1998)). Since the catalytic sites in these enzymes are identical, they are unsuitable for conversion to nicking endonucleases by the present invention. However, experiments with the prototypical Type IIs endonuclease, FokI, indicate that under certain circumstances these homodimerizing enzymes can be coaxed into acting as ‘combinational’ heterodimers (Bitinaite et al., Proc. Natl. Acad. Sci. USA 95:10570-10575 (1998)). As the subunits of such heterodimers differ, they are in this form suitable for conversion to nicking enzymes. Compensating deficiencies may be introduced into a Type IIs and IIf endonucleases that would render the enzymes catalytically inactive, individually as homodimers, but catalytically active, in combination, as heterodimers. Such heterodimers would catalyze DNA-nicking rather than DNA-cleavage.

[0150] Bitinaite et. al. (Proc. Natl. Acad. Sci. USA 95:10570-10575 (1998)) corrupted the dimerization surface of the catalytic domain of R.FokI to prevent homodimer formation, and found that the resulting enzyme, FokI (D483A, R487A), displayed DNA-nicking activity and much reduced DNA-cleaving activity. We can depict such dimerization-defective ‘hemidimers’ as B+C+D−. Kong et al. (U.S. Pat. No. 6,395,523) extended this line of experimentation to other Type IIs endonucleases, and in so-doing created highly active, hemidimeric DNA-nicking enzymes, N.MlyI and N.AlwI that also display much reduced DNA-cleaving activity. When diluted, R.FokI exhibits disproportionately low DNA-cleaving activity due to infrequent homodimer formation. Bitinaite et al. (1998) showed that enzyme activity in these circumstances could be elevated not only by adding more wild-type protein (B+C+D+), but also by adding the separated FokI catalytic domain (C+D+), or a second mutated enzyme, N13Y, that no longer specifically bound to the FokI recognition sequence (B−C+D+). Elevated activity in the former instance derived from wild-type FokI homodimers of composition B+C+D+:B+C+D+, and in the latter instances from combination heterodimers of composition B+C+D+:C+D+, and B+C+D+:B−C+D+, respectively.

[0151] Waugh and Sauer demonstrated that the catalytic site mutants, FokI(D450A), and FokI(D467A), continued to bind to the FokI recognition sequence, but no longer catalyzed DNA strand-hydrolysis (Waugh and Sauer, Proc. Natl. Acad. Sci. USA 90:9596-9600 (1993)). Assuming that these catalysis-deficient enzymes dimerize properly, they can be depicted as B+C−D+. In a 1:1 mixture of a Waugh-type, catalysis-deficient enzyme (e.g. D450A) and a Bitinaite-type binding-deficient enzyme (e.g. N13Y, or the separated catalytic domain), approximately one-half of the dimers that form will be hybrids between the two proteins. In principle, these combinational heterodimers, of composition B+C−D+:B−C+D+, or B+C−D+:C+D+, should bind to DNA through the functional DNA binding-domain of one subunit and catalyze strand-hydrolysis through the functional catalytic domain of the other subunit, under circumstances where neither homodimeric enzyme could. Since the heterodimer in these cases would possess only one catalytic site, it would specifically nick DNA rather than cleave it.

[0152] This approach to creating a nicking endonuclease by combinational heterodimerization has considerable generality. It is by no means certain that best combination of mutations, even in the case of FokI, would be those equivalent to D450A, D467A, N13Y, or the separated catalytic domain, however, and so experimentation with various combinations on a case-by-case basis is to be expected in practice. For FokI, and for other Type IIs endonucleases like it, the following enzymatic scenario can be envisaged: 13 R.FokI (B+C+D+:B+C+D+) GGATGNNNNNNNNNˆ NNNNN . . . active homodimer CCTACNNNNNNNNNNNNNˆ N . . . Mutant 1 (B+C−D+:B+C−D+) Inactive Mutant 2 (B−C+D+:B−C+D+ or Inactive C+D+:C+D+) N.FokI (B+C−D+:B−C+D+ or GGATGNNNNNNNNNˆ NNNNN . . . B+C−D+:C+D+) nicking ‘heterodimer’ CCTACNNNNNNNNNNNNNNN . . .

[0153] (In the above, enzymes of one strand-specificity may be generated.) Type IIf restriction endonucleases (also referred to as ‘Type IV’ endonucleases; (Janulaitis et al., Nucl. Acids Res. 20:6043-6049 (1992)) are organized somewhat differently to Type IIs enzymes, but they too should be amenable to the kinds of manipulations described in this Example for FokI, and appropriate combinations of defective Type IIf enzymes should also manifest sequence-specific, strand-specific, DNA-nicking activities. Experimental investigation of Type IIf enzymes, and comparative analysis of their amino acid sequences, shows them to be combination endonuclease-methyltransferases of some complexity (Janulaitis et al., Nucl. Acids Res. 20:6051-6056 (1992); Kong et al., J. Biol. Chem. 269:683-690 (1994); Sears et al., Nucleic Acids Res. 24:3590-3592 (1996); Piekarowicz et al., J. Mol. Biol. 293:1055-1065 (1999)). The N-terminal 200-250 aa of the proteins comprises a catalytic domain for DNA strand hydrolysis (Rimseliene and Timinskas, Biologija. 1:31-33 (1997); Kong, J. Mol. Biol. 279:823-832 (1998)). The central 350-400 aa comprises a catalytic domain for DNA methylation, and usually resembles the gamma-variety of N6-adenine methyltransferases (Wilson, Methods Enzymol. 216:259-279 (1992)). And the C-terminal 250 aa comprises a sequence-specific DNA-binding domain, and often resembles a single recognition-domain of a Type I endonuclease-methyltransferase specificity-subunit. The complexity of these proteins stems from their combined enzymatic activities; their structural variability (in some enzymes the DNA-binding domain is an integral part of the enzyme, and in others it is separate subunit); and from their organizational variability (some enzymes bind to DNA as monomers, and others bind as dimers or trimers and then form higher-order homodimers between these assemblages in order to catalyze DNA cleavage).

[0154] The simplest Type IIf enzyme, such as Eco57I, can be described as C+D+M+B+ in the monomeric, DNA-bound form, and C+D+M+B+:C+D+M+B+ in the active, dimerized form. Since the catalytic and DNA-binding domains of the Type IIf enzymes are distinct and well separated, discretely inactivating either should present no more of a challenge than would the same for Type IIs enzymes. Combining the resulting two, inactive, enzymes—one deficient in catalysis, the other deficient in DNA-binding—should produce a DNA-nicking activity comparable to that envisaged for FokI: 14 RM.Eco57I active CTGAAGNNNNNNNNNNNNNNNNˆ N . . . homodimer (C+D+M+B+:C+D+M+B+) GACTTCNNNNNNNNNNNNNNˆ NNN . . . N.Eco57I nicking CTGAAGNNNNNNNNNNNNNNNNˆ N . . . ‘heterodimer’ (C−D+M+B+:C+D+M+B− GACTTCNNNNNNNNNNNNNNNNNN . . . or C−D+M+B+:C+D+)

[0155] (In the above sequences, the strand-specificity of the hypothetical nicking enzyme is shown to nick the top strand.)

[0156] The methyltransferase activity of the BcgI endonuclease-methyltransferase can be abolished without loss of DNA-cleavage activity (Kong and Smith, Nucleic Acids Res. 25:3687-3692 (1997); Kong, J. Mol. Biol. 279:823-832 (1998)). The same is probably true for other Type IIf enzymes, and so methylation-deficient mutants of Eco57I could likely be constructed that would mimic ordinary Type IIs endonucleases in function. In the Example above, then, the two defective enzymes that in combination create the nicking activity need not be M+, but either could be M− instead. The binding domain of these enzymes is an extension of the methyltransferase domain, and the two activities are interdependent to such an extent that mutations in either domain would likely abolish both activities. It would probably be rather straightforward, therefore, to make the following combination: C−D+M+B+:C+D+M−B− or C−D+M+B+:C+D+.

[0157] Certain Type IIf enzymes, such as HaeIV, cleave DNA on both sides of their recognition sequences (Piekarowicz et al., J. Mol. Biol. 293:1055-1065 (1999)). These enzymes bind to DNA as homodimers, and then in all likelihood accomplish cleavage by dimerization between the catalytic domains of adjacent homodimers. Combinations of defective mutants of these enzymes, too, could produce DNA nicking activity, although in these cases, since the recognition sequences are symmetric, the nicks would occur on both sides of these sequences, albeit on opposite strands. For HaeIV, the following activities are predicted:

[0158] RM.HaeIV cleaving homomultimer

[0159] (C+D+M+B+:C+D+M+B+):(C+D+M+B+:C+D+M+B+)

[0160] . . . NNNNNNˆ NNNNNNNNGAYNNNNNRTCNNNNNNNNNNNNNˆ N . . .

[0161] . . . Nˆ NNNNNNNNNNNNNCTRNNNNNYAGNNNNNNNNANNNNNN . . .

[0162] Hypothetical N.HaeIV nicking heteromultimers

[0163] (C−D+M+B+:C−M+B+):(C+D+M±B−:C+D+M±B−) or (C−D+M+B+:C−D+M+B+):C+D+

[0164] . . . NNNNNNNNNNNNNNNGAYNNNNNRTCNNNNNNNNNNNNNˆ N . . .

[0165] . . . Nˆ NNNNNNNNNNNNNCTRNNNNNYAGNNNNNNNNNNNNNNN . . .

[0166] (In the above sequences the location of the nick sites are shown downstream of the recognition sequence.)

[0167] Melting of the DNA between the two nicks created by a HaeIV-like nicking enzyme combination would produce DNA fragments with unusually long—37 base in this Example—single-stranded ends of unique, complimentary sequence. Such ends could be useful markers in complex genomes, and might be used to facilitate the proper assembly of such genomes following engineering.

[0168] Type IIs and Type IIf restriction endonucleases that cleave DNA as homodimers can in principle be converted into combinational heterodimers that nick DNA in the following steps:

[0169] 4.1 The gene for the endonuclease is cloned, its nucleotide sequence is determined, and the amino acid sequence of the endonuclease protein is computed.

[0170] 4.2 The amino acid sequence is analyzed for catalytic and structural motifs, and it is compared to the sequences of other endonucleases. The likely functional organization of the enzyme is established.

[0171] 4.3 Mutants are made in the putative DNA-binding domain with the object of discretely abolishing sequence-specific DNA binding, without affecting folding, dimerization, or catalytic ability. The mutations can be targeted or random, and they can include point-mutations, deletions and insertions. Such mutants will likely not harm the host cell in the absence of protective modification. Crude, or partially purified, enzyme is prepared from the mutants, assayed for DNA cleavage activity and shown to be inactive. The desired, B−C+D+, or C+D+ enzyme mutants can be recognized by their ability to enhance the activity of dilute wild-type enzyme preparations.

[0172] 4.4 Mutants are made in the putative catalytic site with the object of discretely abolishing catalysis without affecting protein-folding, DNA-binding, or dimerization. Such mutants will also likely not harm the host cell in the absence of protective modification. Crude, or partially purified, enzyme is prepared from the mutants, assayed for DNA cleavage activity and also shown to be inactive. These enzyme preparations are then mixed with preparations of B−C+D+ or C+D+ enzymes from the previous step and re-assayed, this time for DNA-nicking activity. The desired, B+C−D+ enzymes can be recognized by their ability to nick DNA sequence-specifically in combination with an appropriate B−C+D+ or C+D+ enzyme preparation.

[0173] 4.5 The mutants can be isolated in reverse order if necessary, B+C−D+ enzymes being isolated prior to the isolation of B−C+D+ or C+D+0 enzymes. Once mutants of both types have been isolated, each can be used, in turn, to recover more effective mutants of the other in boot-strap fashion.

Claims

1. A method for converting into a sequence-specific, strand specific and location specific DNA nicking endonuclease, a restriction endonuclease that recognizes an asymmetric DNA sequence, the endonuclease having two catalytic sites and one or more single sequence specific DNA-binding domains, the method comprising:

inactivating one of the catalytic sites of the restriction endonuclease by modifying the nucleotide sequence coding for selected amino acid residues in the catalytic site so as to convert the restriction endonuclease into the nicking endonuclease.

2. A method according to claim 1, wherein the restriction endonuclease has one single sequence specific DNA-binding domain.

3. A method according to claim 1 wherein the restriction endonuclease further comprises a dimerization domain that remains unmodified in the nicking endonuclease.

4. A method according to claim 1, wherein the two catalytic sites of the restriction endonuclease comprise different amino acid sequences.

5. A method according to claim 1, wherein the two catalytic sites of the restriction endonuclease comprise the same amino acid sequence.

6. A method according to claim 2, wherein the restriction endonuclease is selected from BstNB1 and Bbv1.

7. A method according to claim 5, wherein the restriction endonuclease is a dimer, the dimer having a first and a second subunit wherein the first and the second subunit each comprise a sequence specific DNA-binding domain, a catalytic site and a dimerization domain, the method further comprising modifying the first subunit by inactivating the catalytic site and modifying the second subunit by inactivating the DNA-binding domain, the modified dimer forming the nicking endonuclease.

8. A method according to claim 1, wherein the restriction endonuclease prior to modification is a homodimer and the nicking endonuclease is functionally a heterodimer.

9. A method according to claim 7, wherein the restriction endonuclease is a type IIs or type IV restriction endonuclease.

10. A method according to claim 7, wherein the restriction endonuclease is FokI.

11. A method according to claim 1, wherein inactivating one of the catalytic sites further comprises modifying one or more amino acid residues in a predetermined region of one of the catalytic site by site specific mutagenesis of the nucleic acid encoding the residues.

12. A method according to claims 2 or 7, wherein inactivating one of the catalytic sites further comprises: modifying by site specific mutagenesis, a nucleic acid sequences encoding the residues of one or more amino acid residues in a predetermined region of the restriction endonuclease so as to inactivate one of the catalytic sites or the DNA binding domain

13. A method according to claims 2 or 7, wherein inactivating one of the catalytic sites further comprises: modifying by site specific mutagenesis, a nucleic acid sequences encoding the residues of one or more amino acid residues in a predetermined region of the restriction endonuclease so as to inactivate one of the catalytic sites or the DNA binding domain.

14. A method according to claims 1, 2 or 7, wherein inactivating one of the catalytic sites further comprises modifying one or more amino acids in a conserved motif in the DNA encoding the restriction endonuclease, the conserved motif identified through alignment of nucleic acid sequences of related proteins.

15. A method according to claim 14, wherein the conserved motif encodes any of a proline (P), aspartic acid (D), glutamic acid (E) and lysine (K) amino acids in a PD-EXK motif.

16. A method according to claim 1, wherein the restriction endonuclease is selected from AciI, BbvCI, Bpu10I, BsrBI, BssSI, BtrI and GdiII.

17. A method according to claim 3, wherein the restriction endonuclease is selected from BsmI, BsrI and SimI.

18. A method according to claim 3, wherein the restriction endonuclease is selected from BsrDI, BstF5I and BtsI.

19. A method according to claim 3, wherein the restriction endonuclease is selected from BstNBI and BbvI.

20. A method for converting into a sequence-specific, strand specific and location specific DNA nicking endonuclease, a Type II restriction endonuclease that recognizes an asymmetric DNA sequence, the endonuclease being a dimer having a first and second subunit wherein the first and the second subunit each comprise a sequence specific DNA-binding domain, a catalytic site and a dimerization domain, the method comprising: modifying the first subunit by inactivating the catalytic site and modifying the second subunit by inactivating the DNA-binding domain, the modified dimer forming the nicking endonuclease,

21. A nicking endonuclease made according to any of claims 1, 2, 7 and 20.

22. A method of introducing site specific nicks into pre-selected strands of a DNA duplex, comprising mixing double stranded DNA with a nicking endonuclease formed according to any of claims 1, 2, 7 and 20.

23. A method of claim 22, wherein the double stranded DNA is circular supercoiled DNA.

24. A method of converting into a restriction endonuclease, a DNA nicking endonuclease according to claim 21, comprising; adding manganese cation to a digestion buffer containing the nicking endonuclease.

25. A method of inhibiting activity of a nicking endonuclease according to claim 21, comprising: adding calcium to the endonuclease in a digestion buffer.